Aromatic Clusters as Potential Hydrogen Storage Materials

The scientific community is engrossed in the thought of a probable solution to the future energy crisis keeping in mind a better environment-friendly alternative. Although there are many such alternatives, the green hydrogen energy has occupied most of the brilliant minds due to its abundance and numerous production resources. For the advancement of hydrogen economy, Government agencies are funding pertinent research projects. There is an avalanche of molecular systems which are studied by several chemists for storing atomic and molecular hydrogens. The present review on molecular hydrogen storage focuses on all-metal and nonmetal aromatic clusters. In addition to the effect of aromaticity on hydrogen trapping potential of different molecular moieties, the importance of using the conceptual density functional theory based reactivity descriptors is also highlighted. Investigations from our group have been revealing the fact that several aromatic metal clusters, metal doped nonmetal clusters as well as pure nonmetal clusters can serve as potential molecular hydrogen trapping agents. Reported systems include N4Li2, N6Ca2 clusters, Mgn, and Can (n = 8–10) cage-like moieties, B12N12 clathrate, transition metal doped ethylene complexes, M3 + (M = Li, Na) ions, E3-M2 (E = Be, Mg, Al; M = Li, Na, K) clusters, Li3Al4 − ions, Li decorated star-like molecules, BxLiy (x = 3–6; y = 1, 2), Li-doped annular forms, Li-doped borazine derivatives, C12N12 clusters (N4C3H)6Li6 and associated 3-D functional material, cucurbiturils, lithium–phosphorus double-helices. Ni bound C12N12 moieties are also reported recently.


INTRODUCTION
With the ever-increasing utilization of energy, whose primary source has been fossil fuels for so long, the rate of carbon dioxide concentration in the atmosphere is increasing at an alarming rate. With this level of energy demand, fossil fuels will soon be exhausted unless more and more clean fuels are adopted. As of 2019, the International Energy Agency (IEA, 2019) reports the world total energy consumption, of which oil source constitutes 40.4%, followed by 19.7% consumption of electricity, 16.4, 10.4, and 9.5% of natural gas, biofuels, and coal, respectively, and 3.6% constitute other sources of energy. Although 2020 has witnessed a significant reduction in global CO 2 emissions (by 2.4 gigatons) and a decline in the usage of electricity owing to the industrial sector shutdown as a part of COVID-19 restrictions, we are far from reaching the goal of saving the world from collapsing due to over usage of non-renewable energy resources. On the bright side, more and more countries are announcing pledges towards attaining net-zero emissions by the year 2050. To achieve such goals, more and more research projects are being undertaken to search for alternative reliable clean energy sources like hydrogen, nuclear, and efficiently harness the natural resources that we already have in the form of solar, hydro, and wind.
Hydrogen, as a fuel source, is ideal since we have an abundance of it, it causes no emission of harmful gases, and it has a much greater energy content per unit than fossil fuels (World Nuclear Association Website, 2016). The challenges that arise include the conversion of hydrogen from various sources like water and hydrocarbons to its free state, followed by its storage and transportation. The first challenge can be overcome by processes like electrolysis of water and steam reformation of small hydrocarbons, both of which have disadvantages of their own. While the former requires electricity, the latter produces CO 2 as a by-product. Between the two, the electrolysis process seems more preferable since it can be made eco-friendly by using solar, wind, and hydro electricity. Again, there is a downside of higher cost. After the production of free molecular hydrogen, it needs to be stored in an environment with high pressure and very low temperatures. A better way is to adsorb or entrap H 2 in molecular clusters and cages such that the desorption process would also be feasible. Such materials must follow certain standard requirements set by the United States Department of Energy (DOE) (U.S. Department of Energy, USCAR, 2017) to be considered as efficient storage material. In order for a compound to practically act as an effective hydrogen storage material, it must be highly stable, easily available, light weight, inexpensive, showing fast adsorption-desorption kinetics at ambient conditions, can achieve high gravimetric and volumetric hydrogen storage density, and favourable thermodynamic parameters. The appropriate binding energy range depends on the type of hydrogen adsorption on the storage material. For physisorption, it is very small (in the mili-eV range), for chemisorption the value ranges within 2-4 eV, and for the type between physisorption and chemisorption, the binding energy ranges from 0.1 to 0.8 eV. Interaction energy that lies in between that of physisorption and chemisorption is ideal for a better reversible hydrogen storage material.
The aromatic stability of a compound can influence the hydrogen adsorbing power of the molecular motif as described on alkali metal doped hydrocarbons by Srinivasu et al. (2009) Others are discussed in this review. The concept of aromaticity is not yet completely unfolded. It started with being able to explain the reason behind the extra stability in some cyclic hydrocarbons following certain rules prescribed by Hückel, 1931a, Hükel, 1931b, Hückel, 1932. Thorn and Hoffman (1979 then contributed to explaining those of metallabenzene compounds. The accurate measurement of this behaviour is somewhat challenging in the sense that there is no "one-size-fits-all" concept here. Numerous descriptors can measure aromaticity using various techniques but they are highly system-dependent. In this report, we have described the influence of aromatic behaviour on the H 2 trapping potential of various all-metal and nonmetal systems where it is measured in terms of nucleus independent chemical shift (NICS) (Schleyer et al., 1996;Chen et al., 2005) values.

Mg/Ca n and Li/Na 3 + Clusters
The hydrogen storage potential of magnesium and calcium cages (Mg n and Ca n ; n 8-10) along with that of trigonal alkali-metal cationic clusters (Li 3 + and Na 3 + ) were explored by Giri et al. (2011a) with a CDFT approach. The stability of 1H 2 -trapped Mg n and Ca n complexes increases with increase in n. It is, however, interesting to discover that these bare cages could not be Frontiers in Energy Research | www.frontiersin.org November 2021 | Volume 9 | Article 786967 stabilized i.e., before trapping the H 2 molecule within them. Again, in the cases of H 2 -bound Li 3 + and Na 3 + systems, the stability is found to increase with the increase in the number of bound H 2 molecules. Upon binding with the clusters, most of the H 2 retains their molecular identity, except in a few of the nH 2 Li 3 + systems where one of the H 2 molecules dissociates and binds itself in the atomic form. This behaviour is also observed in the case of H 2 -trapped Ca 10 cage ( Figure 1). The planarity of the Li 3 + and Na 3 + clusters remained intact in the H 2 -bound complexes. Further analysis of the changes in the CDFT based reactivity indices, viz., electronegativity (χ) hardness (η) and electrophilicity (ω), revealed a gradual decrease in the ω values with an increase in the number of H 2 molecules.
The NICS (0, 1) values calculated for the top and bottom M 4 rings (M Mg, Ca) of the H 2 -encapsulated Mg n and Ca n cages are found to be negative, indicating the existence of diatropic ring current at the said positions. Similarly, negative NICS zz (0) values for the poly-hydrogenated trigonal clusters also established the aromatic stability with gradual H 2 uptake. Further, from a thermodynamic point of view, the negative values of the reaction energy (ΔE b , kcal/mol) for the sequential H 2 binding on the Li 3 + and Na 3 + systems provide some theoretical justification towards their possible usage as hydrogen storage materials ( Table 1).

Be 3 M 2 , Mg 3 M 2 , and Al 4 M 2 (M Li, Na, K) Clusters
The aromaticity of all-metal Al 4 2− system was previously reported (both theoretically and experimentally) (Li et al., 2001), followed by that of Be 3 2− and Mg 3 2− systems (Kuznetsov and Boldyrev, 2004;Roy and Chattaraj, 2008;Giri et al., 2010). All these systems owe their aromatic behaviour to the presence of delocalized π-electrons. The motivation for using these systems with the alkali metal counterion M + (M Li, Na, and K) (Srinivasu et al., 2012) is backed by the logic that due to high polarization, the M centers will be highly electropositive and hence can serve as a suitable site for hydrogen adsorption. The optimized geometries of the systems under study are provided in Figure 2, and their hydrogen adsorbed structures are depicted in Figure 3. The positive charges carried by the M centers were found to be proportional to the difference in electronegativity values between the two types of metal species in the complex. This difference is lower in Mg 3 M 2 as compared to that in Be 3 M 2 and Al 4 M 2 , which causes a lower charge transfer from M to the Mg 3 ring compared to the others. This, in turn, results in poor H 2 adsorption in the Mg 3 M 2 complexes (interaction energy of −0.5 kcal/mol per H 2 , compared to the same ranging between −1.8 to −3.1 and −2.7 to −3.2 kcal/mol per H 2 for Be 3 M 2 and Al 4 M 2 , respectively). Now, from the perspective of the alkali metals, a high ionic radius and low charge density (i.e., low ionic potential) on the potassium atom results in a very weak interaction with the K atom containing systems (as low as −0.6 kcal/mol per H 2 ). Both Be 3 Li 2 and Be 3 Na 2 are found to adsorb six hydrogen molecules each (three per alkali metal atom), whereas Al 4 Li 2 and Al 4 Na 2 are capable of binding eight molecules each (four H 2 per alkali metal atom). The gravimetric weight percentage of hydrogen adsorption for the former two are 22.64 and 14.12, those for Al 4 Li 2 and Al 4 Na 2 are 11.59 and 9.4, respectively. Evidently, Be 3 Li 2 is preferred to Al 4 Li 2 in terms of the gravimetric density, whereas with respect to the binding energy per H 2 molecule, Al 4 Li 2 is preferred as a better hydrogen storage material. The stability of these hydrogenated systems is established in terms of higher energy gaps (5.17-6.40 eV) between their respective HOMO and LUMO. The variation in total energy (E), χ, η, and ω also serve as good descriptors for judging the reactivity and stability of the molecular species. A uniform decrease in E, and decrease in ω values with increasing H 2 adsorption describes the stabilization of the metal clusters upon hydrogenation. η also shows a gradual increase in most of the cases. The aromaticity of Be 3 M 2 and its hydrogenated species calculated in terms of NICS values at 0, 0.5, and 1 are highly negative and show very low variation with the number of H 2 molecules loaded. For the Al 4 M 2 cluster, however, most of the NICS values are found to be positive, even though it is an established fact that the Al 4 2− species has aromatic stability. The variation of aromaticity with the number of adsorbed H 2 molecule is also found to be quite random and significantly high. Clearly, NICS is not a good indicator for the Al 4 M 2 clusters.

Planar Molecular Stars
Several non-metallic aromatic/anti-aromatic clusters are also explored to understand the influence of aromaticity on their hydrogen adsorbing or trapping potential. Replacing the H atoms from planar hydrocarbon rings like C 4 H 4 , C 5 H 5 − , and C 6 H 6 with Li atoms results in the formation of star-shaped molecular clusters since the Li atoms prefer to bind two adjacent C atoms via a bridging bond (Figure 4) (Giri et al., 2011c). The partial positive charges on the Li centers enable them to readily adsorb H 2 molecules. Each Li atom in C 5 Li 5 − can efficiently bind with only one H 2 (binding energy of −2.1 kcal/mol per H 2 ), whereas those in the neutral clusters (i.e., in C 4 Li 4 and C 6 Li 6 ) can bind up to two H 2 molecules with binding energies ranging within −2.5 to −3.3 kcal/mol per H 2 . A thorough analysis of the NICS-scan (0-5 Å) plots ( Figure 5) for each of the studied species provides useful information regarding the variation in the degree of aromaticity/antiaromaticity with gradual hydrogen loading. The antiaromatic nature of C 4 H 4 is found to be enhanced in its lithium analogue. It further increases in the 4H 2 @C 4 Li 4 system, followed by a decrease in 8H 2 @C 4 Li 4 where the NICS (0) value falls below those of the parent moieties  Figure 5A). Since the C 4 Li 4 species is more likely to exist in the 8H 2 -bound form, it can be arguably stated that hydrogen adsorption on C 4 Li 4 reduces its antiaromatic nature, which is favourable in terms of explaining the stability of these clusters. In the case of the aromatic C 5 H 5 − ring, lithiation increases its aromatic nature as evident from Figure 5B. The 5H 2 -bound C 5 Li 5 − complex follows the same trend as the parent moieties while moving from NICS (4) to NICS (1), i.e., they gradually decrease with decreasing distance from the ring center. However, unlike the parent moieties, 5H 2 @C 5 Li 5 − shows an unexpected rise in NICS (0) value rendering it anti-aromatic. Again, unlike in C 5 -species, the aromaticity reduces in the lithium analogue of C 6 H 6 complex. Although all the species maintain their aromatic behaviour, the degree of aromaticity  initially decreases with hydrogen loading (i.e., in 6H 2 @C 6 H 6 ) with a final increase in the 12H 2 @C 6 H 6 species ( Figure 5C). These plots provide ample evidence regarding the dominant π-ring current in the C 6 species compared to that in the C 5 complexes, which results in an enhanced H 2 -trapping ability in the former.

Li-Doped Annular Systems and 3D Molecular Stars
The H 2 binding ability of a series of lithium ion complexed hydrocarbons (Li + above and below the planar ring), viz., C 6 H 6 , C 10 H 8 , and C 14 H 10 are explored in the same study (Giri et al., 2011c), along with that of an F − ion stabilized tropylium complex. Each Li atom in these systems can store up to four H 2 molecules ( Figure 6A). The same is true for the C 7 H 7 F system, even though, unlike the Li + -complexed hydrocarbons, C 7 H 7 + loses its planarity and hence aromaticity on complexation with F − . Other 3D molecular star-like systems like C 5 Li 7 + (π-aromatic and σ-nonaromatic) and its Si analogue (σ-and π-aromatic) can bind up to a maximum of 21 H 2 molecules (three per Li center) with a gravimetric wt% of 28.0 and 18.3, respectively (Pan et al., 2012;Perez-peralta et al., 2011;Tiznado et al.,2009) (Figure 6B). The Ge analogue has a maximum capacity of 19 H 2 molecules with 9.3 wt% of H 2 (Contreras et al., 2013). It is a well-known fact that temperature and pressure can act as tuning parameters in various adsorption/desorption processes. An interesting study performed on a series of Li clusters (BLi 6 + , O 2 Li 5 + , N 2 Li 7 + , B 2 Li 11 + , OLi 3 + , C 2 Li 9 + , F 2 Li 3 + , and FLi 2 + ) was reported by Pan et al. (2012) where they have achieved enhanced interaction between Li and H 2 by applying an external electric field. By gradually varying the field in the range 0.001-0.005 au, the negative ΔE value increases indicating improved interaction.

Lithium Doped Boron Hydrides
A discussion on hydrogen storage warrants the inclusion of boron hydrides. Metal borohydrides like LiBH 4 and NaBH 4 are reported to be good hydrogen storage materials (Kojima et al., 2002;Züttel et al., 2003;Liang et al., 2010). The effect of aromaticity on the structural preference of both neutral and ionic boron hydrides with the general formula B 3 H n (n 3-9) was reported by Korkin et al. (1995). A similar study (Bandaru et al., 2012) was performed on several boron-lithium complexes with the general formula B x Li y (x 2-6; y 1, 2) where the cationic species show a higher H 2 binding capacity than their neutral counterparts owing to the greater net positive charge on the Li centers in the former. Among various boron hydrides, B 3 H 3 2− can be considered as the iso-electronic boron-analogue of the aromatic cyclopropenium ion (Garratt, 1986;Minkin et al., 1994). A theoretical study on B 3 H 3 2− and its various Li doped species performed by Pan et al. (2011) investigates the H 2 storage process via an interaction that lies somewhere between physisorption and chemisorption. The aromaticity of these systems is evident from the negative values of NICS (0) (ranging within −3.32-−29.38) and NICS (1) (varies from −7.48 to −17.91). The parent moiety itself is capable of adsorbing a maximum of 11 H 2 molecules. Here, the importance of the dispersion interaction is clearly reflected from the interaction energy values obtained before and after including the dispersion corrected terms in the theoretical calculations. For example, the interaction energy per H 2 molecule in the maximum H 2 -loaded systems ranges within −1.2-−1.9 kcal/mol, which changes to a range of −2.6-−5.6 kcal/mol after dispersion correction. Although judging by parameters like the H 2 binding capacity, and gravimetric wt% of H 2 , the B 3 H 3 2system seems to be a preferable choice, it is yet to be experimentally realized. In that regard, B 3 (μ-Li) 3 H 3 + species is a better alternative with high aromatic stability. The H 2 molecules are mostly bound by weak interaction and most of the adsorption process shown here is energonic in nature. Thus an optimal condition for these processes to be feasible would be that of high pressure (P) and low temperature (T). A T-P phase diagram (Figure 7) for the adsorption of 6 H 2 molecules on B 3 (μ-Li) 3 H 3 + depicts the variation of free energy change with the change in T and P, where black markers indicate regions of favourable adsorption and red markers indicate that of favourable desorption.

Planar N 4 2− and N 6 4− Rings
A study (Duley et al., 2011) reported on the aromaticity and hydrogen storage in the inorganic analogues of cyclobutadiene and benzene rings, i.e., N 4 2− and N 6 4− reveals the anti-aromatic and aromatic natures, respectively. The NICS-scan plots of all four systems are provided in Figure 8 where it can be clearly seen that the NICS variation of N 6 2− follows a similar trend as that of the benzene, whereas this is not the case for N 4 2− and cyclobutadiene. The otherwise antiaromatic N 4 2− ring shows aromaticity at a distance of 1 Å above the molecular plane, and gradually decreases away from the ring. Thus, a positive NICS (0) value and a negative NICS (1) value makes it σ-antiaromatic and π-aromatic. These anionic clusters are stabilized by complexation with lithium and calcium counterions to form N 4 Li 2 and N 6 Ca 2 (optimized structures provided in Figure 9). Each Li center is found to bind with 3.2.5 C 12 N 12 , B 12 N 12 , and Ni-Decorated C 12 N 12 Cages C 12 N 12 cage can have three possible isomers, viz., C 12 N 12 -A, C 12 N 12 -B, and C 12 N 12 -C with the optimized structures having D 6d , C 3 , and C 2 point group symmetries, respectively ( Figure 10A) (Mondal et al., 2013). Unlike isomers A and B, C is an open cage structure and hence is the most stable of the lot. Although the NICS (0) for all the isomers are found to be negative, isomers A and B do not obey the Spherical Aromaticity rule [2(N+1) 2 π-electrons], and isomer C does not follow the Open-Shell Spherical Aromaticity rule [(2N 2 + 2N + 1) π-electrons]. Although, the hydrogen may be expected to bind at three possible sites: N, C, and bridging C and N, the minimum energy structures are obtained only for the N-site adsorption. All the isomers are capable to bind 12 H 2 molecules providing a gravimetric wt% of 7.2 ( Figure 10B). These clusters can have potential applications as high energy density materials (HEDMs) since they have high heat of formation values (ΔH 0 f ). These ΔH 0 f values (1747.759, 1758.442, and 1813.864 kcal/mol for isomers A, B, and C, respectively) are calculated using the following isodesmic reaction: C 12 N 12 + 12H 2 C CH 2 + 3NH 3 → 9C 4 H 4 NH + 3H 2 N−NH 2 The Ni-decorated analogue (Jana et al., 2020) of the cage can have three possible binding modes: bridging between C and N (denoted as X CN isomer), between 2 C atoms (X CC isomer), or between 2 N atoms (X NN isomer) ( Figure 11A). Higher numbers of Ni decoration (up to four) in the N-(μ-Ni)-N bridging mode reveals the variation in aromaticity and H 2 -binding potential with the number of Ni atoms ( Figure 11B). The reason for selecting N-(μ-Ni)-N bridging mode for this purpose is the highest amount of positive charge on the Ni center of the X NN isomer. The isodesmic reactions used to determine the ΔH 0 f of Ni-C 12 N 12 (4,239.470 kcal/mol) and 4Ni-C 12 N 12 (4,520.528 kcal/mol) are as follows: Ni−C 12 N 12 + 12H 2 C CH 2 + 3NH 3 → 9C 4 H 4 NH + 3H 2 N − NH 2 + Ni 4Ni−C 12 N 12 + 12H 2 C CH 2 + 3NH 3 → 9C 4 H 4 NH + 3H 2 N −NH 2 + 4Ni Each Ni center on C 12 N 12 is capable to adsorb up to three H 2 molecules ( Figure 11C). In the single Ni-doped clusters, the aromaticity increases in the order X CN < X CC < X NN , and in the case of multiple Ni-doping, the aromaticity increases uniformly with the number of Ni atoms.
The boron analogue of the C 12 N 12 cage, i.e., B 12 N 12 can trap hydrogen both endohedrally and exohedrally . A total of 12 hydrogen molecules can be adsorbed in this cage system, the interaction energy per H 2 molecule varies from −11.5 to −12 kcal/mol. The change in the reactivity descriptors with the gradual hydrogen uptake validates the CDFT based electronic structure principles. The aromaticity in terms of NICS (0) values of all the nH 2 @B 12 N 12 (n 0-12) species are plotted and presented in Figure 12. It is evident that among all the hydrogen-loaded species, the four, eight, and twelve numbers of H 2 loaded clusters have high aromatic stability.
In this report, we have described the influence of aromatic behaviour on the H 2 trapping potential of various all-metal and nonmetal systems where it is measured in terms of nucleus independent chemical shift (NICS) values. The effect of aromaticity is not uniform throughout all systems. In some systems, the variation of aromaticity with the number of adsorbed H 2 molecule is quite uniform and systemic, like in the cases of C 4 Li 4 , C 5 Li 5 − , and C 6 Li 6 rings, N 4 2− and N 6 4− rings, the nNi-decorated C 12 N 12 clusters. While for others, like that in the Al 4 M 2 clusters, it is found to be quite random and significantly high. It deserves a careful scrutiny.

CONCLUSION
In this review, we have covered several case studies predicting all-metal and non-metal aromatic clusters with potential applications in hydrogen storage and highlighted the effect of aromaticity on the hydrogen uptake potential. The aromatic stabilization present in these clusters enables them to be applicable as promising building blocks for nanomaterials. The NICS-scan plots turn out to be very helpful in understanding the degree and trend in the variation of the aromatic/anti-aromatic nature. The presence of an electropositive center in the molecular clusters provides a FIGURE 11 | Optimized structures of (A) C 12 N 12 and its three Ni-bound isomers, (B) nNi-decorated C 12 N 12 , (C) H 2 -trapped single Ni-bound C 12 N 12 clusters at ωB97X-D/Def-TZVP level. suitable site for the H 2 molecules to undergo a dipole-induced dipole interaction for adsorption. Temperature and pressure are good tuning parameters in controlling the adsorption/ desorption process. In this context, a T-P phase diagram can provide proper guidance as to which temperature and pressure range could be ideal to undergo a favourable adsorption process.
We have also highlighted the use of external electric field in tuning the H 2 loading and release processes and hence overall improving the H 2 trapping capacity of the cluster in question. For future research, the search for chemical frameworks with large surface area and high porosity can be made for better reversible hydrogen storage via physisorption having interaction energy closer to that of chemisorption, with a target of improving the gravimetric and volumetric storage capacity, favourable rates of adsorption and desorption, operating temperatures, and thermal conductivity. Molecular frameworks offering higher surface area to volume ratio can increase volumetric capacity. The binding energies can be increased by tuning the pore volume, and the thermal conductivity can be improved with the help of new and developed heat exchangers.

AUTHOR CONTRIBUTIONS
PC came up with the concept and design of the review, wrote the abstract, reviewed the final manuscript. RP contributed towards the literature survey, writing the manuscript. All authors contributed to manuscript revision, read, and approved the submitted version.